Free course

Prompt Engineering

Write prompts that get reliable, reusable results from ChatGPT, Claude, Gemini or any AI assistant. Six short lessons built around your own work, each with an exercise and a test that opens the next one.

Last updated: November 2, 2026

What is prompt engineering?

Prompt engineering is the skill of writing instructions that get an AI assistant to produce the result you need, reliably, on the first or second try. It is less about secret phrases and more about briefing: giving the assistant the task, the context, the examples and the format a capable new colleague would need to do the job well.

It matters because the same assistant can be excellent or useless depending on how it is asked. In a well known Harvard Business School study with Boston Consulting Group, 385 consultants using GPT-4 finished 12.2% more tasks, 25.1% faster and with more than 40% higher quality on tasks the AI handled well. On a task outside what the AI could do, the consultants who relied on it were 19 percentage points less likely to get the right answer. Good prompting is how you get the first result, and good checking is how you avoid the second.

What you will be able to do

• Write a prompt with the four parts that matter: role, task, context and format.
• Brief an assistant like a colleague, so its first draft needs fewer fixes.
• Use examples and structure to get output in exactly the shape you need.
• Break a large task into a chain of smaller prompts that check and improve each other.
• Test a prompt on several inputs and catch made-up facts before they reach anyone else.
• Build a personal library of prompts you reuse every week.

What each lesson covers

1. How AI assistants read your prompt. What an assistant knows and does not know when you ask it something, and the four parts of a good prompt. Exercise: score your current prompts for three real tasks.
2. Brief it like a colleague. Context, audience, purpose, constraints and the reason behind each instruction. Exercise: rewrite your first prompt as a full briefing.
3. Show the format, give examples. Examples, delimiters and output formats, and where to put long documents. Exercise: build a prompt with examples for a repeated task.
4. Break big tasks into steps. Prompt chains, plans before answers, and self-review, plus how reasoning models change the rules. Exercise: turn a multi-step job into a chain.
5. Check before you trust. Why assistants invent facts, how to test a prompt on several inputs, and a checking routine for anything that matters. Exercise: test your best prompt on five inputs.
6. Build your prompt library. Reusable templates, custom instructions, projects and custom assistants, and what not to paste into a chat. Exercise: build a library of five prompts and measure the change.

Who it is for

Anyone who uses an AI assistant at work, or wants to: marketers, managers, founders, freelancers, analysts, writers, support and sales teams. No technical knowledge is needed. The lessons work with ChatGPT, Claude, Gemini, Copilot or any similar assistant.

What you need

• A free account on at least one AI assistant.
• Three tasks from your own work that you would like an assistant to help with.
• About 30 minutes per lesson, including the exercise, at your own pace.
• The course workbook, a PDF with one sheet per exercise, which you download in lesson 1.

Questions about the course

Do I need to learn special phrases or tricks?

No. Most “magic” phrases stop working or never worked. The course teaches the habits that hold across assistants and model versions: clear tasks, real context, examples, structure, steps and checking.

Does it work with ChatGPT, Claude and Gemini?

Yes. The techniques come from the official guidance of OpenAI, Anthropic and Google, and from research on how people use these tools at work. Where an assistant behaves differently, the lesson says so.

Is prompt engineering still useful now that models are smarter?

Yes, but it has changed. Newer models need fewer tricks and more context. They still cannot know your audience, your standards or your data unless you tell them, and they still make things up. Briefing and checking matter more than ever.

Is the course up to date?

Every lesson shows the date it was last updated, and we revise the lessons when the tools or the guidance change. Figures always name their source and date.

What do I get at the end?

The course is free, and so are the workbook and the tests. When you pass the test of every lesson, you can get a certificate of completion with your name, a public verification link and a button to add it to LinkedIn. The certificate costs $9, or is included in the Annual Pass for $20 a year. It shows that you completed the course; it is not an accredited qualification.